Now that installation of PTC is mandated by Federal law, I continue to be surprised by the lack of knowledge (even among railfans) about what it is and what it does. in part, I think, this is because FRA chose to write a functional specification rather than a prescriptive regulation -- there are several different ways of meeting the statutory requirements. I'll attempt to briefly summarize a complex topic (I've been working on PTC-related issues on and off since 1988).
1) PTC is required by statute on all tracks carrying more than 5 MGT of freight traffic, on all tracks carrying passenger trains -- with some exceptions -- and on all tracks over which any volume of TIH is moved. The statute only covers Class I railroads; Class II and Class III carriers are exempted, and passenger trains operating on those carriers may also be exempted, on an individual basis, by the Secretary of Transportation.
2) There are two distinct "flavors" of PTC. One is the ETMS-type system well described in the video link posted earlier on this thread. This is a "vehicle-centric" system; vehicle borne computers determine train position and speed and enforce authorities transmitted directly from the central office. The only equipment on the wayside is radio repeaters and "wayside interface units" that monitor switch position and grade crossing activation. WIUs are very cheap and simple devices. The other flavor of PTC is what I call "wayside-centric". In this arrangement, most communications are between wayside devices (either track-mounted transponders, as in Amtrak's ACSES, or wayside "beacons" in ITCS) and trains. The wayside devices transmit authorities and grade crossing and switch data, as well as maximum speeds. There need not, in these systems, be any communication between trains and the central office. In this respect, ACSES and ITCS are like conventional cab signal systems.
3) The information generated by a vehicle-centric PTC system can be used for a host of other functions, such as dispatch planning, schedule adherance, locomotive health monitoring, status of en-route pickups and set-outs, etc. Some people (and I am one) believe that the availability of real-time train location data will enable railroads to realize substantial operating efficiencies. The railroads deny this. But for example, how much more efficient will track maintenance become when a foreman knows, to the minute, when the next train will show up?
4) To say ETMS-type systems are "GPS based" does not mean they rely exclusively on GPS, and that the whole railroad will come to a halt if the GPS signal is lost. Position is also determined by odometers, accelerometers, and triangulation from radio towers (just as cell phones do). And of course, once an authority is issued, the train can proceed to its limit even if GPS is lost (or even if radio contact is lost). No new authorities can be issued, of course, until communications are restored. Also remember that all Class I railroads have encoded their tracks into GIS databases, so if you monitor switches, you know which way the train went at the last switch, so there is no problem in discriminating between two tracks on a double-track line -- unless the train is not on the track, in which case you have other problems.
5) Since much of the cost in vehicle-centric systems is either on the locos or in the central office, the current carping over the percentage of the system that must be equipped is a canard. Since all locomotives will have to have the equipment, and all tracks on Class I railroads also have voice radio coverage, the most rational thing to do is equip the entire network. Digital radios can be sited at the same locations as existing radio base stations. Switch position indicators are already required (they can be visible targets) even on unsignaled track, and WIUs are cheap.
6) Similar objections concerning the difficulty of calculating a braking algorithm are also red herrings. In 1989, on the Iron Range of Minnesota, I watched a dispatcher remotely intervene and stop a 22,000 ton ore train. It stopped right where it needed to, just short of a red home signal. If BN could do that in 1988 with an 8086 processor and 4800 baud digital data link, I have trouble believing it can't be done today with processors and data links much faster than those. AAR has conjured up some worst case scenario of a train with non-functional dynamic brakes (which loco health monitoring would detect in a PTC system) and 15% inoperative train brakes (for which FRA would fine a railroad for letting a train like that leave a terminal). There are already "adaptive" braking algorithms that "learn" how a particular train handles, and I have no doubt that they'll do a better job than many human engineers. I speak from the perspective of having run a lot of different train simulation packages over the last 20 years.
So despite all the moaning and grumbling, I'm confident, first, that PTC will work (I can't see any insuperable technical difficulties -- after all, BN had a working prototype more than 20 years ago), and second, that it will turn out to produce substantial operating benefits for railroads. So let's quit complaining and go do it. The law gives the railroads five years.
1) PTC is required by statute on all tracks carrying more than 5 MGT of freight traffic, on all tracks carrying passenger trains -- with some exceptions -- and on all tracks over which any volume of TIH is moved. The statute only covers Class I railroads; Class II and Class III carriers are exempted, and passenger trains operating on those carriers may also be exempted, on an individual basis, by the Secretary of Transportation.
2) There are two distinct "flavors" of PTC. One is the ETMS-type system well described in the video link posted earlier on this thread. This is a "vehicle-centric" system; vehicle borne computers determine train position and speed and enforce authorities transmitted directly from the central office. The only equipment on the wayside is radio repeaters and "wayside interface units" that monitor switch position and grade crossing activation. WIUs are very cheap and simple devices. The other flavor of PTC is what I call "wayside-centric". In this arrangement, most communications are between wayside devices (either track-mounted transponders, as in Amtrak's ACSES, or wayside "beacons" in ITCS) and trains. The wayside devices transmit authorities and grade crossing and switch data, as well as maximum speeds. There need not, in these systems, be any communication between trains and the central office. In this respect, ACSES and ITCS are like conventional cab signal systems.
3) The information generated by a vehicle-centric PTC system can be used for a host of other functions, such as dispatch planning, schedule adherance, locomotive health monitoring, status of en-route pickups and set-outs, etc. Some people (and I am one) believe that the availability of real-time train location data will enable railroads to realize substantial operating efficiencies. The railroads deny this. But for example, how much more efficient will track maintenance become when a foreman knows, to the minute, when the next train will show up?
4) To say ETMS-type systems are "GPS based" does not mean they rely exclusively on GPS, and that the whole railroad will come to a halt if the GPS signal is lost. Position is also determined by odometers, accelerometers, and triangulation from radio towers (just as cell phones do). And of course, once an authority is issued, the train can proceed to its limit even if GPS is lost (or even if radio contact is lost). No new authorities can be issued, of course, until communications are restored. Also remember that all Class I railroads have encoded their tracks into GIS databases, so if you monitor switches, you know which way the train went at the last switch, so there is no problem in discriminating between two tracks on a double-track line -- unless the train is not on the track, in which case you have other problems.
5) Since much of the cost in vehicle-centric systems is either on the locos or in the central office, the current carping over the percentage of the system that must be equipped is a canard. Since all locomotives will have to have the equipment, and all tracks on Class I railroads also have voice radio coverage, the most rational thing to do is equip the entire network. Digital radios can be sited at the same locations as existing radio base stations. Switch position indicators are already required (they can be visible targets) even on unsignaled track, and WIUs are cheap.
6) Similar objections concerning the difficulty of calculating a braking algorithm are also red herrings. In 1989, on the Iron Range of Minnesota, I watched a dispatcher remotely intervene and stop a 22,000 ton ore train. It stopped right where it needed to, just short of a red home signal. If BN could do that in 1988 with an 8086 processor and 4800 baud digital data link, I have trouble believing it can't be done today with processors and data links much faster than those. AAR has conjured up some worst case scenario of a train with non-functional dynamic brakes (which loco health monitoring would detect in a PTC system) and 15% inoperative train brakes (for which FRA would fine a railroad for letting a train like that leave a terminal). There are already "adaptive" braking algorithms that "learn" how a particular train handles, and I have no doubt that they'll do a better job than many human engineers. I speak from the perspective of having run a lot of different train simulation packages over the last 20 years.
So despite all the moaning and grumbling, I'm confident, first, that PTC will work (I can't see any insuperable technical difficulties -- after all, BN had a working prototype more than 20 years ago), and second, that it will turn out to produce substantial operating benefits for railroads. So let's quit complaining and go do it. The law gives the railroads five years.
Randy Resor, aka "NellieBly" passed away on November 1, 2013. We honor his memory and his devotion to railroading at railroad.net.